基于方向局部二值模式的虹膜识别方法  

An Iris Recognition Method Based on D-LBP

作　　者：黄敏[1] 刘文超[1] 任宗华[1] 常化文[1] 

机构地区：[1]郑州轻工业学院计算机与通信工程学院,河南郑州450002

出　　处：《湖北民族学院学报（自然科学版）》2016年第4期415-419,共5页Journal of Hubei Minzu University(Natural Science Edition)

基　　金：国家自然科学基金青年科学基金项目(61401404)

摘　　要：虹膜识别技术作为最简单最精确的人体生物识别技术,越来越受人们的重视.目前,CS-LBP算子由于它的高效性和灰度不变性,在虹膜识别领域取得了较好的识别效果.但它忽略了中心像素对虹膜纹理的影响,因此提出了基于D-LBP的虹膜识别方法.它将周围邻近的像素和中心像素进行比较,重新定义了特征描述码的计算方法.该方法在降低特征维数的同时,很好的解决了CS-LBP算法中心像素丢失的问题.实验结果表明:与LBP和CS-LBP方法相比,提出的方法具有更好的识别效果.Iris recognition technology,as the simplest and most accurate human biometric technology,has attracted more and more attention.At present,the CS-LBP operator has achieved a good recognition result in the field of iris recognition because of its high efficiency and gray-scale invariance.But it didn’t con-sider the influence of the center pixel on the iris texture. Therefore,this paper proposes an effective im-proved method based on D-LBP for iris recognition. It compares the neighboring pixels with the center pixels,and redefines the calculation method of the signature description codes. This method not only re-duces the dimension,but also solves the problem of the loss of the center pixel of CS-LBP algorithm.And the experiments show that the method proposed by the paper has a better recognition effect compared with LBP and CS-LBP methods.

关 键 词：虹膜识别 中心对称局部二值模式(CS-LBP) 方向局部二值模式(D-LBP) 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]